Detecting groups of coherent voxels in functional MRI data using spectral analysis and replicator dynamics

Area of Science:

• Neuroimaging
• Computational Neuroscience
• Data Analysis

Background:

• Functional magnetic resonance imaging (fMRI) generates time series data reflecting brain activity.
• Analyzing the relationships between fMRI time series is crucial for understanding brain connectivity.
• Existing methods like the number of coherent voxels (NCV) have limitations in accurately identifying coherent voxel groups.

Purpose of the Study:

• To investigate the relationship between human brain fMRI time series.
• To combine fMRI spectral analysis with replicator dynamics for improved analysis.
• To develop a reliable method for detecting groups of maximally coherent voxels.

Main Methods:

• Bivariate spectral coherence was used to analyze simulated and real fMRI time courses.
• Coherence values were organized into matrices representing voxel relationships.
• Replicator dynamics were applied to detect groups of maximally coherent voxels.
• Results were compared against the number of coherent voxels (NCV) method.

Main Results:

• The NCV method's performance is critically dependent on an arbitrarily chosen threshold.
• Higher NCV thresholds failed to detect all coherently related voxels in simulations.
• Replicator dynamics successfully identified the entire group of coherent voxels across all simulations.
• The replicator process proved more robust and accurate than NCV.

Conclusions:

• Replicator dynamics applied to spectral matrices offer a reliable approach for identifying maximally coherent voxel groups.
• This method effectively determines groups where all voxels exhibit high coherence with each other.
• The replicator dynamics method is parameter-free and avoids the need for a priori reference voxel selection, unlike NCV.